<<

J. Limnol., 2013; 72(s2): 1-17 GEOLOGICAL HISTORY DOI: 10.4081/jlimnol.2013.s2.e1

The palaeogeography of and Wallacea since the Late Jurassic

Robert HALL*

Southeast Research Group, Department of Sciences, Royal Holloway University of London, Egham, Surrey TW20 0EX, UK *Corresponding author: [email protected]

ABSTRACT The continental core of Southeast (SE) Asia, Sundaland, was assembled from Gondwana fragments by the Early Mesozoic. Con- tinental blocks rifted from in the Jurassic [South West (SW) , East -West -], and the Woyla intra- oceanic arc of , were added to Sundaland in the Cretaceous. These fragments probably included emergent areas and could have carried a terrestrial flora and fauna. Sarawak, the offshore Luconia-Dangerous Grounds areas, and Palawan include Asian con- tinental material. These probably represent a wide accretionary zone at the Asia-Pacific boundary, which was an active continental margin until the mid Cretaceous. Subduction ceased around Sundaland in the Late Cretaceous, and from about 80 Ma most of Sun- daland was emergent, physically connected to Asia, but separated by deep oceans from India and Australia. India moved rapidly north during the Late Cretaceous and Early Cenozoic but there is no evidence that it made contact with SE Asia prior to collision with Asia. One or more arc-India collisions during the Eocene may have preceded India-Asiaonly collision. The arcs could have provided dispersal pathways from India into SE Asia before final suturing of the two . During the Late Cretaceous and Early Cenozoic there was no significant subduction beneath Sumatra, Java and Borneo. At about 45 Ma Australia began to move north, subduction resumed and there was widespread rifting within Sundaland. During the Paleogene east and north Borneo were largely submerged, the Makassar Straits became a wide marine barrier within Sundaland, and West Sulawesi was separated from Sundaland but included land. By the Early Miocene the proto-South China Sea had been eliminated useby subduction leading to emergence of land in central Borneo, Sabah and Palawan. Australia-SE Asia collision began, eliminating the former deep ocean separating the two continents, and forming the now known as Wallacea. The microplate or terrane concept of slicing fragments from followed by multiple collisions in Wallacea is implausible. Neogene subduction drove extension and fragmentation of Wallacea that caused both subsidence of deep marine basins and elevation of land; topography and bathymetry changed very rapidly, especially during the Pliocene, but the detailed palaeogeography of this region remains uncertain.

Key words: SE Asia, , subduction, land, sea.

Received: February 2013. Accepted: April 2013. commercial INTRODUCTION DISCUSSION This paper aims to give a brief overview of the late Mesozoic to Early Cenozoic Mesozoic and Cenozoic palaeogeography of the South- Non The core of SE Asia (Fig. 1), Sundaland, was initially east (SE) Asian region, primarily south of Indochina and assembled in the Late Palaeozoic and Early Mesozoic by Thailand. It summarises the main geological events which are discussed in much more detail elsewhere (e.g. Hall, addition of continental fragments carried from Gondwana 1996, 2002, 2011, 2012a) and here the focus is primarily (Fig. 2) to SE Asia (Metcalfe, 2011). From the Jurassic on the palaeogeography illustrated by maps that show the onwards much of Indochina southwards through the Thai- main features of the region during the Cenozoic. Discus- Malay Peninsula and parts of the present as sion of the many problems in making these palaeogeo- far south as Sumatra was an emergent area of land, prob- graphical sketches can be found in Hall (1998, 2001) and ably largely surrounded by subduction zones, implying other accounts of palaeogeography and geology can be that the margins included volcanoes and mountains. The found in Hall (2009, 2012b). I have attempted to keep the limited geological record that is preserved suggests that number of references to the minimum possible most of the interior of this region was the site of deposi- but all these papers include much larger bibliographies tion of terrestrial clastic sediments during the Jurassic and with more complete discussion of the geological Early Cretaceous (Abdullah, 2009; Racey and Goodall, issues, and several are accompanied by tectonic recon- 2009; Clements et al., 2011). struction animations which are available at http://searg. More continental blocks were added in the Early and rhul.ac.uk/current_research/plate_tectonics/ early Late Cretaceous (Fig. 2) to form what is now much 2 R. Hall of Borneo, West Sulawesi and Java. The fragments had been inferred from nearby unmetamorphosed limestones rifted from Australia in the Late Jurassic and their colli- of Permo-Carboniferous age with of Cathaysian sions ended subduction around the SE Asia promontory characteristics (Fontaine, 1990). There are also Triassic at about 90 Ma. The Woyla intra-oceanic arc of Sumatra sedimentary rocks in Sarawak containing a flora was also added to Sundaland at this time. East Java and with Asian affinities (Vozenin-Serra, 1977). These rocks West Sulawesi are now known to be underlain by Precam- had therefore been interpreted as a Palaeozoic metamor- brian Australian continental crust rather than Mesozoic phic basement overlain by sedimentary rocks (Hutchison, accreted crust (Smyth et al., 2007). South West (SW) Bor- 2005) accreted to Sundaland during the Sarawak orogeny neo was thought to include Palaeozoic metamorphic (Hutchison, 1996) and I had suggested these formed part rocks, and has been regarded as part of a Sunda shield, of a Luconia-Dangerous Grounds block (Hall, 2012a). but our recent work shows that these supposed ancient However, having recently seen these rocks in the field and rocks are largely metamorphosed Cretaceous vol- reconsidered the evidence for the Sarawak orogeny (Hall, canogenic rocks, although we have a few hints of a Pre- 2012a; Hall and Sevastjanova, 2012), I consider it more cambrian basement beneath (Davies et al., 2012). These likely that the metamorphic rocks are not a Palaeozoic discoveries raise the possibility that the Australian frag- basement but instead that Sarawak, the offshore Luconia- ments may have brought with them some terrestrial faunal Dangerous Grounds areas, and Palawan include Asian and floral elements from Gondwana in the Cretaceous. continental material as well as ophiolitic/arc rocks which The SW Borneo rocks have often been correlated with probably represent a wide accretionary zone at the Asia- metamorphic rocks of Sarawak which were also thought Pacific boundary, which wasonly an active continental margin to be Palaeozoic. These too are undated; their age has until the mid Cretaceous. use

commercial

Non

Fig. 1. Simplified of Wallacea, Sundaland and surrounding . The 200 m bathymetric contour has been used to define the edge of the Asian margin, Sunda shelf and the Sahul-Arafura shelf. Black filled triangles are volcanoes. Lines with small triangles represent subduction zones with triangles on the upper plate. D and S mark the Dent and Semporna peninsulas of eastern Sabah, and ST is the Seram trough. Sundaland-Wallacea palaeogeography 3

Clements et al. (2011) suggested that the termination of than a South American Andean mar- subduction in the early Late Cretaceous contributed to wide- gin, with locally emergent islands separated from an in- spread uplift of Sundaland, marked by a major regional un- terior Sundaland by shallow seas. To the south conformity. Rocks below the unconformity are Cretaceous and east were deep water oceanic areas of the Pacific and or older, and above are Eocene or younger, representing a Indian oceans. time gap of more than 80 million years (Clements et al., 2011). It is reasonable to suppose that most of Sundaland Cenozoic: Paleogene was an emergent region, and mountainous, supported by the Two geological events are likely to have influenced the presence of Laurasian conifer pollen (Muller, 1968) in distribution of land and sea, and terrestrial biogeographic Sarawak sandstones, from the Late Cretaceous to Middle Eocene. The few rocks of this age that are preserved, notably patterns, in the Sundaland region in the Early Cenozoic. in Sarawak and NW Kalimantan, are predominantly terres- These are the collision of India and Asia, and the resump- trial clastic sediments deposited by rivers, except at the ex- tion of subduction around Sundaland at about 45 Ma. treme margins of Sundaland in Sarawak, Sabah, West India carried a faunal and floral cargo so the age of Sulawesi, and probably offshore East Java, where there are connections with SE Asia which would allow dispersals poorly dated deep marine clastic deposits. A prolonged pe- is very important, but a clear-cut answer to many ques- riod of emergence and widespread reworking of older rocks tions is not easy to provide. Ali and Aitchison (2008) have is suggested by the character of the oldest sediments in the suggested that India could have made contact with the many Cenozoic sedimentary basins that formed throughout Sumatran part of Sundaland in the Paleocene, before the Sundaland when sedimentation resumed in the Eocene. main India-Asia collision.only I see no geological evidence to They are typically quartz-rich conglomerates and sand- support this suggestion, but there is evidence for an India- stones, very mature in terms of grain shape and composition, arc collision at this time, and it is plausible that the arc suggesting multiple episodes of recycling. could have provided a dispersal route into western Sun- At the end of the Cretaceous subduction resumed be- dalanduse long before there were continuous land connec- neath West Sulawesi, and continued through the Pale- tions from India to Asia. The timing of India-Asia ocene (Fig. 3), and there was probably a continental collision remains a controversial issue. There are two margin volcanic arc, probably more like the present-day main views. The first view is that India-Asia -

commercial

Non

Fig. 2. Sundaland and Wallacea. The heavy dashed lines show the limit of the Sundaland continent at different stages in the Cretaceous and Cenozoic. Wallacea is shaded in light grey. Its western boundary is most commonly drawn through the Makassar straits and across the Celebes sea at Wallace’s line and the eastern boundary approximately follows the Australian-New Guinea shelf edge. Darker grey shaded areas are Cenozoic oceanic basins. 4 R. Hall continent collision occurred at about 50 Ma with the im- molecular evidence to date appearances of organisms that plication that from this time it would have been possible require complete terrestrial connections. for organisms to cross from one continent to another with Until recently, most reconstructions assumed uninter- no marine barriers. The second view is that India-Asia rupted subduction around Sundaland from the Cretaceous continent-continent collision did not occur before 35 Ma into the Cenozoic. However, there is almost no subduc- and until that time there would have been marine barriers tion-related igneous record at the Sundaland margins, in to dispersal. The geological arguments are many and com- contrast to the period before 90 Ma and the period after plex and, as so often is the case, much of the evidence can 45 Ma. During the Late Cretaceous and Early Cenozoic be interpreted in different ways. Studies of closures of there was some oceanic spreading south of Australia, but other marine passages, such as the Panama and Indone- Australia remained close to Antarctica. I have therefore sian throughflow regions, suggest that a single age of col- suggested that subduction beneath Sumatra, Java and Bor- lision may be a misconception. The term collision is used neo ceased at about 90 Ma and resumed around Sunda- in multiple ways, from the timing of first impact to com- land at about 45 Ma as Australia and Antarctica separated plete suturing, the detailed record of land and sea is prob- and Australia began to move north at a faster rate. Sub- ably never completely reconstructable, and the duction initiation was associated with widespread changes requirement of geographic connection can be very differ- on land and rifting at the Sundaland margins, and by the ent for different organisms, ranging from wind dispersal Middle Eocene (Fig. 4) it appears that much of Sundaland of seeds to continuous freshwater habitats that can allow was topographically low. Rivers fed clastic sediments to transit on foot. This is one area where the biologists may the Sundaland margins ofonly Sumatra, Java, SE Borneo and be able to help the if they could use fossil and Sarawak, implying a still elevated interior which included use

commercial

Non

Fig. 3. Palaeogeography at 60 Ma - Paleocene. Much of Sundaland was emergent and there was probably a relatively high interior. There was a volcanic arc at the eastern edge of Sundaland in West Sulawesi and Sumba. Sundaland-Wallacea palaeogeography 5 the Malay Peninsula and the Schwaner Mountains of SW vated region supplying sediment to Sunda shelf basins, Borneo. Thermochronological studies indicate that subsi- and the Schwaner mountains of SW Borneo was another. dence offshore in the Gulf of Thailand and Malay Basin Parts of the present Java sea such as the Karimunjawa and was associated with uplift of the Malay Peninsula. Despite Bawean arches (Smyth et al., 2003, 2008), which occupy rapid subsidence most basins were not bathymetrically the area between Borneo and Java, may also have been deep features [probably less than 200 m, based on Shoup elevated and provided sediment. In the area west of the et al. (2012)] and contain fluviatile and largely marginal present Meratus Mountains, known as the Barito basin, marine deposits. there was a wide river system, where peats and fluvial and The palaeogeography varied with eustatic sea level estuarine sediments were deposited, which appears to fluctuations, but from the Eocene to Early Miocene have flowed north during the Eocene from the present (Figs. 4-7) most of western Sundaland was terrestrial with Java sea (Witts et al., 2012) towards the Makassar straits, deposition in sedimentary basins dominated by fluviatile with limestones deposited during periods of higher sea input, with many often large freshwater lakes. The palaeo- level. Clastic sediment was also transported into the North geographic maps presented in Figs. 4-9 incorporate de- Makassar straits from central or west Borneo. Although tailed outlines of the numerous freshwater and brackish the marine area of eastern Borneo increased in size and lakes and inland seas which have been mapped by Shoup became deeper, most of the rest of Sundaland was terres- et al. (2012) and Morley and Morley (2013) from the in- trial with locally small areas of higher elevation. tegration of an extensive seismic and biostratigraphic The part of eastern known to biologists as database. It is likely that the Malay peninsula was an ele- Wallacea dates inonly part from the Eocene (Fig. 4). Rifting use

commercial

Non

Fig. 4. Palaeogeography at 40 Ma - Late Eocene. Subduction resumed around much of Sundaland in the Middle Eocene and new sed- imentary basins were filled mainly by terrestrial sediments. The Makassar straits had become a significant marine gap at the eastern edge of Sundaland, isolating West Sulawesi. 6 R. Hall of the SE Sundaland margin (Hall and Morley, 2004), areas. SE Sundaland from South Sulawesi to East Java probably caused by the development of new subduction was largely an area of shallow water carbonate deposition. zones, separated West Sulawesi from Borneo and created At the southern margin of Sundaland between the the Makassar straits. Initially this was a wide but partly Eocene and Early Miocene (Fig. 6) were volcanic arcs. terrestrial gap (alluvial plains crossed by rivers with peaty The volcanoes mainly formed islands rather than contin- swamps), and partly shallow marine (shallow marine car- uous and extensive areas of land. In Sumatra volcanic ac- bonates and clastic sediments), with elevated ridges par- tivity became widespread from the Middle Eocene. The allel to the rift which were locally submerged and capped Eocene arc was initially constructed on the edge of Sun- by shallow water carbonates. By the Oligocene the daland with some volcanic centres in a terrestrial setting Makassar straits had become a wide and deep marine bar- and others forming islands on a coastal plain. Later marine rier within Sundaland (Fig. 5). West Sulawesi was physi- transgressions left the volcanic Barisan mountains as a cally isolated from the rest of Sundaland but included land chain of large islands south of the elevated Malay penin- which continued to supply sediment to adjacent offshore sula by the Early Miocene (Fig. 7). In Java there was a se-

only use

commercial

Non

Fig. 5. Palaeogeography at 30 Ma - Mid Oligocene. Although much of Sundaland was emergent it is likely that topography was signif- icantly lower than earlier in the Cenozoic. Rivers carried recycled clastic sediments to internal basins and the continental margins. On the Sunda shelf there were large freshwater lakes, not linked to the ocean. Land remained in West Sulawesi. Details of Sunda shelf freshwater lakes from Morley and Morley (2013). Sundaland-Wallacea palaeogeography 7 ries of small volcanic islands south of the Sundaland canic arc of Sabah-Cagayan, leading to emergence, or in- coast. From the Eocene to Early Miocene the proto-South crease in area, of land in central Borneo, Sabah and China sea was subducted southwards beneath northern Palawan. Australia began to collide with SE Sundaland in Borneo where there was volcanic activity. Deep water Sulawesi about 23 million years ago (Figs. 6 and 7), effec- areas at the active subduction margin received quartz-rich tively closing the former deep ocean separating the two con- sediment from the Malay-Thai peninsula and the tinents. These collisions led to mountain building in Borneo, Schwaner mountains, implying eastward-flowing rivers Sulawesi, and the Banda arc. In addition, the arrival of arcs from granitic source areas. Some material was derived from the Pacific led to the emergence of islands in east In- from emergent areas in Sabah where ophiolitic rocks form donesia. They changed the palaeogeography of the region the basement. in a significant way by enlarging the area of land in Wal- lacea, but also leading to an increase in the areas of shallow Cenozoic: Neogene seas both in Wallacea and further west in Sundaland. By the Early Miocene (Fig. 7) the proto-South China For many years geologists have interpreted the com- sea had been eliminated by subduction, resulting in collision plex geology of Wallacea in terms of collision. Within the of continental crust of the South China margin with the vol- context of convergence of Australia and SE Asia many

only use

commercial

Non

Fig. 6. Palaeogeography at 25 Ma - Late Oligocene. On the Sunda shelf brackish lakes were intermittently connected to the sea. The Sula spur was about to collide with the Sulawesi North Arm volcanic arc. Details of Sunda shelf fresh and brackish water lakes from Morley and Morley (2013). 8 R. Hall features have been interpreted as the result of multiple has been driven by extension, related to subduction, and minor collisions of moving tectonic fragments, often a much more complex picture of vertical movements, and called terranes. The geological terrane concept has be- changing land-sea patterns, is emerging. come popular with biogeographers as the model appears to account for many complex distribution patterns. How- Sundaland ever, in Wallacea and New Guinea ideas are changing, and Almost all of southern Sundaland was flat south of the biologists need to exercise caution in applying over-sim- area that today forms the Central Borneo Mountains ple geological solutions. The microplate or terrane con- (Figs. 7 and 8). In SW Borneo there is no Cenozoic record cept of slicing fragments from New Guinea followed by preserved, but Cretaceous granites and metamorphic multiple collisions in Wallacea appears increasingly im- rocks now exposed in the Schwaner mountains imply a plausible. Fragmentation has occurred but much of this prolonged period of emergence and erosion. Most of

only use

commercial

Non

Fig. 7. Palaeogeography at 20 Ma - Early Miocene. There was a marine incursion onto the Sunda shelf. Borneo became an important source of clastic sediments which were carried into deep offshore basins to the north, east and southeast. There were extensive areas of carbonate deposition on wide shallow shelves around Sundaland. After collision in Sulawesi there was uplift and the first links to Aus- tralia, although there was no continuous land connection. Much of Wallacea between Sulawesi and the Bird’s Head was the site of shallow marine carbonate deposition with some areas of land which are difficult to define. Details of Sunda shelf marine embayments and lakes from Morley and Morley (2013). Sundaland-Wallacea palaeogeography 9 southern Borneo probably remained close to sea level. To Further north, the Dangerous Grounds are a rifted mi- the east of the present-day Schwaner mountains, and to crocontinental fragment from the South China margin, the south across the present Java sea, there are Eocene to capped by locally emergent carbonate reefs, shoals and Miocene terrestrial sediments that represent emergence atolls. Parts of this block, and particularly the area that during intervals of low eustatic sea level and shallow ma- eventually became Palawan, was land as it drifted south rine limestones that record intervals of higher sea level. between the Eocene and Early Miocene; other parts were Across the East Java sea to South Sulawesi was the long- submerged but were very shallow marine areas where car- lived Eocene to Miocene Paternoster-Tonasa carbonate bonates were deposited (Figs. 5-7). Subduction of the platform (Figs. 4-7). All these areas are still largely unde- proto-South China sea ended in the Early Miocene when formed with the exception of SE Borneo where the nar- the Dangerous Grounds block underthrust northern Bor- row and elongate Meratus mountains ridge (Fig. 9), which neo causing deformation and uplift (Fig. 7). This signifi- represents a reactivated Cretaceous suture, probably rose cantly increased the land area of Palawan and Borneo and between the Late Miocene and Pleistocene (Witts et al., formed highlands in the interior of Borneo. Large rivers 2011; Witts, 2012). flowed outwards as Borneo rose and the land area contin-

only use

commercial

Non

Fig. 8. Palaeogeography at 15 Ma - Mid Miocene. From about 15 Ma the Java subduction zone began to rollback into the Banda em- bayment. This caused major extension in Sulawesi and began to fragment the Sula spur. Details of Sunda shelf marine embayments from Morley and Morley (2013). 10 R. Hall ued to increase through the Neogene. Pulses of uplift and Palawan seems to have collapsed and become smaller in expansion of land area occurred throughout the Neogene area. This is probably linked to extension related to sub- with the most dramatic example being mount Kinabalu in duction of the Celebes sea beneath south Sabah and the Sabah where recent dating has shown Mio-Pliocene rapid Sulu arc. This subduction produced a Middle and Late uplift, recorded by a 7-8 Ma granite (Cottam et al., 2010) Miocene volcanic arc in the Dent and Semporna peninsu- now exposed at 4 km above sea level. Kinabalu is one of las of south Sabah and the Sulu islands (Figs. 9 and 10), the few mountains between New Guinea and the Hi- which although lacking permanent continuous land could malayas that was provably capped by ice during the Pleis- have permitted connections between Borneo and the tocene (Hope, 2004). It is one of several mountains in via intermittently emergent volcanic islands. Sabah that have become unusually high in the few In contrast, as Borneo became emergent, the area to million years and it is likely that the summit was ice-cov- the north and west records gradual flooding of the Sunda ered and ice-free several times in the last 2 million years. shelf accompanied by disappearance of freshwater lakes, As Borneo became higher and larger in land area brackish embayments and a gradual increase in the size

only use

commercial

Non

Fig. 9. Palaeogeography at 10 Ma - Late Miocene. Borneo was now a large elevated area. Subduction rollback was well underway and extension in Sulawesi was accompanied by volcanic activity, subsidence of Bone gulf and oceanic crust formation in the North , and renewed elevation on land. Wallacea probably had a much more complex palaeogeography than shown. There was no slicing of fragments from the Bird’s Head and fragmentation of the Sula spur was the result of extension driven by subduction rollback into the Banda embayment. Sundaland-Wallacea palaeogeography 11 of the marine area extending from the South China sea response to region-wide Sundaland deformation following (Figs. 7-9). Eustatic sea level fluctuations mean that Australian collision in east Indonesia. Subduction-related Sunda shelf coastlines and marine areas fluctuated con- deformation, reflecting abrupt changes in the age of the siderably with time as documented offshore using seismic subducting crust may also have contributed and biostratigraphic data (Shoup et al., 2012; Morley and to elevation of Sumatra and its forearc. From Nias and Morley, 2013). Siberut the forearc south of the Sumatran coast is ex- Sumatra is underlain by continental crust and like the tremely shallow and there is almost no deep basin between Sunda shelf has been above or close to sea level during the the forearc high and the coast, as there is to the NW and Cenozoic. Since the Middle Miocene the Barisan moun- SE. Several of the large islands, such as Nias and Siberut, tains rose and widened (Figs. 9 and 10), largely due to vol- were probably connected at times to the Sumatran main- canic activity and strike-slip faulting, and partly in land although this is less likely for the islands further SE.

only use

commercial

Non

Fig. 10. Palaeogeography at 5 Ma - Early Pliocene. Oceanic spreading in the North Banda sea had ceased but had begun in the South Banda sea and probably the sea. Significant and rapid changes in topography began in Sulawesi, as rollback commenced at the North Sulawesi trench causing widespread extension of North and East Sulawesi. This led to subsidence of Gorontalo bay from a shallow marine carbonate area to present depths of up to 2 km, and further subsidence of Bone bay. Subsidence was accompanied by significant elevation on land. Uplift was about to begin in as the Banda volcanic arc collided with the Australian continental mar- gin. Seram and Timor are the two largest islands in the Banda region to have emerged since 5 Ma. 12 R. Hall

In contrast to Sumatra, Java became the large island build-ups, and further north in West Sulawesi there is little of today more abruptly and more recently. For most of the indication of mountains and no major deformation. The Paleogene, northern Java and the Java sea was a shallow nature of the boundary between West and East Sulawesi shelf that was close to sea level and intermittently emer- is obscure. The Early and Middle Miocene record is very gent. South of the shelf there were prolonged periods of incomplete, although there are some poorly dated terres- activity in a chain of volcanic islands, except in the Mid- trial and marginal marine sediments, probably because dle Miocene when there was widespread shallow marine much of East and SE Sulawesi was land (Fig. 7). carbonate deposition and volcanic activity declined. The After the collision it seems unlikely there was a con- volcanic arc moved abruptly northwards at about 7 Ma tinuous terrestrial connection from Sulawesi to Australia after arrival at the trench of a buoyant basaltic volcanic via the Bird’s Head, as in many parts of the former Sula province on the , causing widespread thrusting spur and in the Bird’s Head there are widespread shallow throughout Java leading to the emergence of most of West marine carbonates. However, in the Banggai-Sula islands Java. Unusual K-rich volcanoes (for example, Muriah and Cretaceous-Paleocene rocks are overlain in the west by Ringgit in East Java) erupted near the edge of the Sunda flat-lying Eocene to Recent shallow marine carbonates shelf during this interval and normal arc volcanism re- with a slight angular unconformity, and further east where sumed only in the last 1 to 2 million years, which broadly there are no carbonates the Mesozoic sequence has been coincides with the emergence of East Java (Fig. 10) that eroded to expose a basement window which may have has made Java the large elongate island of today. been elevated in the early Cenozoic. It is possible that parts of the islands, and onlypossibly other parts of the Sula Wallacea spur, could have become emergent during the Eocene and In Wallacea topography and bathymetry changed very remained as land until the present. rapidly during the Neogene (Figs. 7-10). This is reflected There was an important change in the Middle Miocene in the complex passageway for water which moves from at about 15 Mause (Fig. 8) when widespread extension and the Pacific to the Indian ocean, which probably had local major subsidence began. This was caused by subduction and global climatic consequences, and is a result of a com- rollback into the Banda oceanic embayment. Subduction plex tectonic history. A popular geological model for Wal- can be caused by convergence between two plates, one of lacea suggests that small fragments of continental crust which subducts beneath the other (Fig. 11A), but alterna- were sliced from the Bird’s Head of New Guinea during tively can be the result of a plate sinking into the mantle the Neogene, moved west along a left-lateral strike-slip under the influence of gravity (Fig. 11B). As the plate zone, and collided with Sulawesi. This concept originated sinks, the subduction zone moves ocean-ward, in a with Hamilton (1979) and has been incorporated in many process that has been described as subduction/slab/hinge tectonic models and reconstructions. The model has rollback, or hinge/trench retreat (Hamilton, 2007). The proved popular with biogeographers as this scenario may rollback induces extension in the upper plate to fill the provide a mechanism for dispersal of organismscommercial from space created (Fig. 12). Such extension can cause subsi- Australia into SE Asia. Our recent work in eastern Indone- dence and uplift, and both occurred in Sulawesi. sia has caused us to doubt it. Instead, we now suggest that Extension occurred in several phases. The results of there was a collision in the Early Miocene between the these different extensional episodes were formation of the Sundaland margin of West andNon North Sulawesi and an deep oceanic basins of the Banda sea and the inter-arm Australian continental promontory, the Sula spur. The non-oceanic basins of Gorontalo bay and Bone bay. The promontory extended west from New Guinea on the north North Banda sea formed between 12.5 and 7 Ma (Fig. 9) side of an oceanic Banda embayment and had formed dur- and the South Banda sea between 6 and 2 Ma (Fig. 10). ing Jurassic rifting of the Australian margin. It was later Both inter-arm bays contain sedimentary sequences that fragmented by extension, not by slicing, as explained include extensive shallow marine carbonates, showing below, and the fragments of continental crust scattered that before extension they were close to sea level. Bone around the Banda Arc are almost all the result of this ex- bay subsided in several phases during Banda subduction tensional mechanism. rollback. Major extension was also caused by develop- The Sula spur was the first part of the Australian con- ment of the North Sulawesi subduction zone which tinent to make contact with the Sundaland margin formed at about 5 Ma. Subsidence of Gorontalo bay, to (Figs. 6 and 7) as shown in an animated reconstruction by its present depths of up to 2 km, was the result of this sub- Spakman and Hall (2010). Collision caused deformation duction rollback. that elevated the eastern parts of Sulawesi and produced We can infer that there were quite large and possibly a widespread unconformity between Neogene and pre- long-lived areas of land in Sulawesi that were, or re- Neogene rocks. However, in South Sulawesi shallow ma- mained, elevated as rollback occurred. These are indicated rine carbonate deposition continued, as dispersed by clastic sediments deposited after collision often called Sundaland-Wallacea palaeogeography 13

Celebes Molasse (because of their supposed similarity to tains expose deep crustal rocks intruded by young granites the molasse of the Swiss Alps). The Celebes Molasse is throughout West Sulawesi as the upper crust was stripped thought to have been deposited in the Early to Late off. Much of this sediment has been carried west into the Miocene, implying uplift and erosion, but most of the Makassar straits. The marine gap that existed since the rocks assigned to it are undated, and the best dated are Eocene has narrowed in the last 5 Ma as Borneo has Late Miocene and Pliocene. The inter-arm bays of emerged and Sulawesi has risen, but the central part of Gorontalo and Bone contain thick sediment sequences the Makassar straits may have become deeper in response that could have been eroded during the Early to Late to gravity-driven movements of material from Sulawesi. Miocene from a mountainous region in East, Central and Almost all the non-volcanic islands of the Outer SE Sulawesi, and in many parts of Central and SE Su- Banda arc (Fig. 13) have emerged in the last 3 Ma. As the lawesi there are metamorphic rocks suggesting significant subduction zone rolled back into the Banda embayment erosional removal of upper crust cover rocks. Laterites in the inner volcanic arc was active and moved with it. At parts of East Sulawesi also suggest prolonged periods of about 4 Ma the volcanic arc collided with the southern tropical weathering of a land surface. Present knowledge passive Australian continental margin of the Banda em- allows little more than the statement that there was an el- bayment in East Timor. In Timor and Sumba the arc-con- evated landmass in Sulawesi during much of the Miocene. tinent collision was marked by rapid uplift which moved Most of the present relief of Sulawesi (Fig. 13) seems sedimentary rocks deposited at depths of several kilome- to have been created since the Late Miocene and in the tres below sea level to their present positions of more than past 5 Ma (Fig. 10) there has been a major increase in land one kilometre aboveonly sea level. Islands such as Savu, Roti area and a significant change in elevation. There was and Tanimbar have emerged even more recently. Volcanic clearly a major increase in output of clastic sediment at activity ceased in and Alor in the mid Pliocene by the same time. In Gorontalo bay spectacular very young about 2 million years ago although it continued in the very and rapid subsidence is recorded by numerous carbonate younguse islands of the eastern arc from Damar to Banda. Is- reefs now found at water depths between 1 and 2 km. lands have formed at different times since the Late Pliocene alluvial fan deposits on islands in Gorontalo bay Miocene, and many parts of Wallacea have emerged at are separated from their equivalents in the East Arm by very high rates, for example in Timor Audley-Charles water depths up to 1.5 km. Seismic lines across Bone bay (1986) estimated average rates of uplift of 1.5 km/Myr. show similar thick sediments and subsidence history. As Other large islands such as Seram, , Sumba and parts these areas subsided from close to sea level to 1 to 2 km of Sulawesi probably rose at similar rates, but as they did depths the mountains close by rose to their present eleva- so, basins like the South Banda sea and Weber deep have tions of 2 to 3 km. Deep valleys incised into steep moun- subsided. commercial

Non

Fig. 11. Subduction can be viewed in two ways. A) A conventional view of subduction in which one plate is moving towards a stationary upper plate and is subducted beneath it. Magmas are produced at the volcanic arc as water is released from the subducted plate and its sedimentary cover, causing melting of the mantle wedge above the subducted plate. In this model the trench and arc remain in a fixed position. B) An alternative view of subduction in which the subducted plate falls into the mantle under the influence of gravity. Sub- duction is maintained by rollback of the subduction hinge which results in the movement of the trench and arc in an oceanward direction. Subduction can be maintained without any convergence between the plates. The upper plate must extend and thin, and new mantle must rise into the wedge, to allow rollback to continue. 14 R. Hall

At the eastern edge of Wallacea are the (Figs. 9 and 10). The modern Halmahera arc is a rela- and Sangihe volcanic arcs which are currently colliding. tively recent arrival in Wallacea. It is constructed on Both these arcs formed during the Neogene. The Sangihe older intra-oceanic arcs formed in the Pacific which arc can be traced from Sulawesi northwards into the were part of the Philippine sea plate. Before 25 Ma, the Philippines. In biogeographic terms it is within Wallacea Philippines-Halmahera arc was an intra-oceanic Pacific but in geological terms is arguably part of Neogene Sun- arc, formed above a north-dipping subduction zone as daland. It was constructed on Eocene oceanic crust of Australia moved north. These arcs collided with northern the Celebes sea at the edge of Sundaland, but in the Neo- Australia at about 25 Ma and then moved west along the gene linked the North Arm of Sulawesi which had pre- New Guinea margin north of the left-lateral Sorong fault viously collided with the Sula spur to the Philippines zone (Figs. 9 and 10). Like the islands of the Banda re-

only use

commercial

Non

Fig. 12. Cartoons to illustrate how extension of the upper plate may occur during subduction rollback. A) The initial situation in which there is subduction beneath continental crust, simplified into upper and lower parts, and mantle. Extension is accomplished by the de- velopment of major faults that cut through the entire crust. B) As rollback proceeds a major detachment fault cuts through the crust causing part of the lower crust to be exhumed. Thinning and inflow of mantle below the stretched crust cause uplift of the thinned lower crust and subsidence of the thinned upper crust. Melting may occur in the thinned region due to increased heatflow as hot mantle rises. C) If extension continues the crust may be completely broken, allowing mantle to reach the surface or leading to formation of new ocean crust between stretched remnants of continental crust. Both uplift and subsidence accompany extension. Sundaland-Wallacea palaeogeography 15 gion they are probably larger areas of land now than at carried from the Pacific to the Indian ocean by the Indone- any time in the past. sian throughflow passes by this route (Godfrey, 1996; Major changes in sea level during the Pleistocene Gordon, 2005; Tillinger, 2011). Thus, as sea level fell cur- caused intermittent emergence of the shelf areas of Sunda rent velocities probably increased. and Australia (Fig. 13); modern bathymetric maps and the Even for the Sunda shelf the method of using modern history of global sea level can be used to infer the distri- bathymetry to infer the positions of coastlines is probably bution of land and sea (Voris, 2000). At times the islands valid only for the Pleistocene, and for longer periods it is around the Sunda shelf were connected and the entire doubtful whether eustatic changes in sea level can be re- shelf was emergent. However, even at times of lowest sea liably obtained by stratigraphic analysis of offshore seis- level during Pleistocene glacial intervals the Makassar mic data (Moucha et al., 2008). For Wallacea the problem straits were never narrower than about 75 km and must is even greater. Glacially-drive sea level change would have been a formidable barrier since most of the water have had little effect on Wallacea where shelves surround-

only use

commercial

Non

Fig. 13. Palaeogeography at the present day for comparison with the other palaeogeographical maps. Topography is from merged global bathymetry and Shuttle radar topography mission (SRTM) data (Sandwell and Smith, 2009). Volcanoes are from Siebert and Simkin (2002). The edge of the shallow shelves is drawn at the 120 m bathymetric contour, which is almost identical in position to the 200 m contour used in Fig. 1. The shallow shelves are the area that would have been emergent during the approximately 20,000 years ago. Highlands are areas above 500 m. The map underestimates the areas of carbonates on the shelves. 16 R. Hall ing islands are narrow, and shelf edges very steep. Tec- Malaysia. University of Malaya and Geol. Soc. Malaysia ed. tonic change has been significant on geologically short Ali JR, Aitchison JC, 2008. Gondwana to Asia: , time scales. In many parts of Wallacea, vertical move- paleogeography and the biological connectivity of the Indian ments (both up and down) up to 1.5 km/Ma have been sub-continent from the Middle Jurassic through latest Eocene (166-35 Ma). Earth-Sci. Rev. 88:145-166. recorded and similar rates can be estimated in the Banda Audley-Charles MG, 1986. Rates of Neogene and region and Sulawesi which indicate the possibility of sig- tectonic movements in the Southern Banda Arc based on mi- nificant palaeogeographical change within time periods cropalaeontology. J. Geol. Soc. London 143:161-175. of less than a million years. The geography of the Wal- Clements B, Burgess P, Hall R, 2011. Subsidence and uplift by lacea has changed very significantly in the last 5 Ma, and slab-related mantle dynamics: A driving mechanism for the the overall trend has been a significant increase in land Late Cretaceous and Cenozoic of Continental area, and an increase in elevation (Fig. 13). SE Asia? Geol. Soc. London Spec. Publ. 355:37-51. Cottam MA, Hall R, Sperber C, Armstrong R, 2010. Pulsed em- CONCLUSIONS placement of the Mount Kinabalu Granite, North Borneo. J. Geol. Soc. London 167:49-60. We are still far from a complete understanding of the Davies L, Hall R, Armstrong R, 2012. Cretaceous crust beneath geology of Sundaland and Wallacea, still less the links to SW Borneo: U-Pb dating of zircons from metamorphic and palaeogeography, ocean-atmosphere circulation and cli- granitic rocks. Available from: http://fallmeeting.agu.org/ mate which have influenced biological change, biogeo- 2012/eposters/eposter/t43e-2714/ Fontaine H, 1990. The Terbat formation of Sarawak (Malaysia): graphical patterns and . The maps in this paper a very peculiar limestone,only p. 173-181. In: H. Fontaine (ed.), are still very broad-brush attempts at showing palaeo- Ten years of CCOP research on the Pre-Tertiary of East Asia. geography. The palaeogeography of the Sunda shelf is CCOP Technical Publication Bangkok. now becoming clearer from offshore seismic data and dat- Godfrey JS, 1996. The effect of the Indonesian Throughflow on ing of sedimentary sequences, and because tectonic ef- ocean circulationuse and heat exchange with the atmosphere: a fects are less important there than in eastern Indonesia. review. J. Geophys. Res. 101:12217-12238. Wallacea remains a challenge because of the speed of tec- Gordon AL, 2005. of the Indonesian Seas and tonic change and our incomplete knowledge of its conse- their Throughflow. Oceanography 18:14-27. quences. Mountains have disappeared and risen, deep Hall R, 1996. Reconstructing Cenozoic SE Asia. In: R. Hall and D.J. Blundell (eds.), Tectonic evolution of SE Asia. Geol. waterways have closed and opened, and ephemeral is- Soc. Spec. Publ. 106:153-184. lands have provided connections. Rapid vertical move- Hall R, 1998. The plate tectonics of Cenozoic SE Asia and the ments may mean there was more land than previously distribution of land and sea, p. 99-131. In: R. Hall and J.D. expected and perhaps more short-lived stepping stones. Holloway (eds.), and geological evolution of Geologists have most difficulty with the terrestrial record SE Asia. Backhuys Publ. which is typically incomplete and poorly dated. Biologists Hall R, 2001. Cenozoic reconstructions of SE Asia and the SW can help us with reliable information on evolutioncommercial and dis- Pacific: changing patterns of land and sea, p. 35-56. In: I. Met- tributions of plants and animals. calfe, J.M.B. Smith, M. Morwood and I.D. Davidson (eds.), Faunal and floral migrations and evolution in SE Asia-Aus- tralasia. A.A. Balkema (Swets and Zeitlinger Publ.). ACKNOWLEDGMENTS Hall R, 2002. Cenozoic geological and plate tectonic evolution Work by the SE Asia ResearchNon Group has been funded of SE Asia and the SW Pacific: computer-based reconstruc- over many years by a consortium of oil companies. In ad- tions, model and animations. J. Asian Earth Sci. 20:353-434. dition, the SE Asia Research Group has received support Hall R, 2009. SE Asia’s changing palaeogeography. Blumea 54:148-161. at times from the University of London Central Research Hall R, 2011. Australia-SE Asia collision: plate tectonics and Fund, the Natural Environment Research Council, and the crustal flow. Geol. Soc. Spec. Publ. 355:75-109. Royal Society. I thank Pusat Survei Geologi Bandung, Hall R, 2012a. Late Jurassic-Cenozoic reconstructions of the In- Lemigas, Indonesian Institute of Sciences, and Institut donesian region and the Indian Ocean. 570- Teknologi Bandung, for assistance and many colleagues, 571:1-41. friends and students in the UK, and SE Asia for Hall R, 2012b. Sundaland and Wallacea: geology, plate tectonics help and discussion. I am especially grateful to Bob Mor- and palaeogeography, p. 32-78. In: D.J. Gower, K.G. John- ley and Duncan Witts for assistance with the palaeogeo- son, J.E. Richardson, B.R. Rosen, L. Rüber and S.T. Williams (eds.), Biotic evolution and environmental change graphical maps. I thank Tony Barber and Bob Morley for in . The Systematics Association, Cambridge review comments on the manuscript. University Press. Hall R, Morley CK, 2004. Sundaland Basins, p. 55-85. In: P. REFERENCES Clift, P. Wang, W. Kuhnt and D.E. Hayes (eds.), Continent- Abdullah NT, 2009. Mesozoic , p. 87-131. In: C.S. Ocean interactions within the East Asian marginal Seas. Hutchison and D.N.K. Tan (eds.), Geology of peninsular Union ed. Sundaland-Wallacea palaeogeography 17

Hall R, Sevastjanova I, 2012. Australian crust in Indonesia. Aust. catalog of Holocene volcanoes and their eruptions. Available J. Earth Sci. 59:827-844. from: http://www.volcano.si.edu Hamilton WB, 1979. Tectonics of the Indonesian region 1078. Smyth H, Hall R, Hamilton J, Kinny P, 2003. Volcanic origin of US Government Printing Office, Washington DC, USA. quartz-rich sediments in East Java. Proc. 29th Ann. Conv. In- Hamilton WB, 2007. Driving mechanism and 3-D circulation of donesian Petroleum Ass., Jakarta, 541-559 p. plate tectonics. Geol. S. Am. S. 433:1-25. Smyth HR, Hamilton PJ, Hall R, Kinny PD, 2007. The deep Hope GS, 2004. Glaciation of Malaysia and Indonesia, exclud- crust beneath island arcs: inherited zircons reveal a Gond- ing New Guinea. Elsevier, London: 388 pp. wana continental fragment beneath East Java, Indonesia. Hutchison CS, 1996. The ‘Rajang Accretionary Prism’ and Earth Planet. Sc. Lett. 258:269-282. ‘Lupar Line’ problem of Borneo. Geol. Soc. Spec. Publ. Smyth HR, Hall R, Nichols GJ, 2008. Significant volcanic con- 106:247-261. tribution to some quartz-rich sandstones, East Java, Indone- Hutchison CS, 2005. Geology of North-West Borneo. Elsevier, sia. J. Sedimt. Res. 78:335-356. Amsterdam: 444 pp. Spakman W, Hall R, 2010. Surface deformation and slab-mantle Metcalfe I, 2011. Palaeozoic-Mesozoic History of SE Asia. interaction during Banda Arc subduction rollback. Nat. Geol. Soc. Spec. Publ. 355:7-35. Geosci. 3:562-566. Morley RJ, Morley HP, 2013. Mid Cenozoic freshwater wet- Tillinger D, 2011. Physical oceanography of the present day In- lands of the Sunda region. J. Limnol. 72(Suppl.2):e2. donesian Throughflow. Geol. Soc. Spec. Publ. 355:267-281. Moucha R, Forte AM, Mitrovica JX, Rowley DB, Quéré S, Sim- Tongkul F, 1991. Tectonic evolution of Sabah, Malaysia. J. mons NA, Grand SP, 2008. Dynamic topography and long- Southe. Asian Earth 6:395-406. term sea-level variations: there is no such thing as a stable Voris HK, 2000. Maps of Pleistocene sea levels in Southeast continental platform. Earth Planet. Sc. Lett. 271:101-108. Asia: shorelines, river systems and time durations. J. Bio- Muller J, 1968. of the Pedawan and Plateau Sand- geogr. 27:1153-1167.only stone Formations (Cretaceous - Eocene) in Sarawak, Vozenin-Serra C, 1977. [Contribution à l’étude de la Paléoflore Malaysia. Micropaleontology 14:1-37. du Sud-est Asiatique (Cambodge, Laos, Vietnam)]. [PhD Racey A, Goodall JGS, 2009. Palynology and stratigraphy of Thesis in French]. University of Paris ed., France. the Mesozoic Khorat group of NE Thailand. Geol. Soc. Witts D,use 2012. Stratigraphy and sediment : the Barito Spec. Publ. 315:67-81. Basin, Southeast Kalimantan, Indonesia. PhD Thesis. Royal Sandwell DT, Smith WHF, 2009. Global marine gravity from Holloway University of London ed., London. retracked Geosat and ERS-1 altimetry: ridge segmentation Witts D, Hall R, Morley R, BouDagher-Fadel MK, 2011. versus spreading rate. J. Geophys. Res. 114:B01411. Stratigraphy and sediment provenance, Barito Basin, South- Shoup RC, Morley RJ, Swiecicki T, Clark S, 2012. Tectono- east Kalimantan. Proc. 35th Ann. Conv. Indonesian Petro- stratigraphic framework and Tertiary paleogeography of leum Ass. IPA11-G-054, pp. 1-18. Southeast Asia: Gulf of Thailand to South Vietnam Shelf. Witts D, Hall R, Nichols G, Morley R, 2012. A new deposi- Available from: http://www.searchanddiscovery.com/docu- tional and provenance model for the Tanjung formation, ments/2012/30246shoup/ndx_shoup.pdf Barito Basin, SE Kalimantan, Indonesia. J. Asian Earth Siebert L, Simkin T 2002. Volcanoes of the world: an illustrated Sci. 56:77-104. commercial

Non